Update src/streamlit_app.py

src/streamlit_app.py

@@ -13,6 +13,7 @@ import hdbscan
 
 # Database file path
 DB_PATH = '/data/steampolis.duckdb'
+DEFAULT_BASE_URL = 'https://huggingface.co/spaces/npc0/SteamPolis/'
 
 # Initialize database tables if they don't exist
 def initialize_database():
@@ -97,6 +98,210 @@ def initialize_database():
         if 'init_con' in locals() and init_con:
             init_con.close()
 
+
+import random # Import random for generating votes
+
+def add_example_topic(topic_id, topic_title, topic_description, comments_list):
+    """
+    Adds an example topic, its comments, and example cluster votes to the database.
+
+    Args:
+        topic_id (str): The unique ID for the topic.
+        topic_title (str): The title of the topic.
+        topic_description (str): The description of the topic.
+        comments_list (list): A list of strings, where each string is a comment.
+    """
+    con = None
+    try:
+        con = duckdb.connect(database=DB_PATH)
+
+        # Insert the topic
+        con.execute("""
+            INSERT INTO topics (id, title, description)
+            VALUES (?, ?, ?)
+            ON CONFLICT (id) DO NOTHING
+        """, [topic_id, topic_title, topic_description])
+
+        # --- Add Cluster Users ---
+        # Create users who will cast votes, separate from comment authors
+        num_users_per_cluster = 5
+        cluster1_users = [] # e.g., Pro-Tech supporters
+        cluster2_users = [] # e.g., Anti-Tech skeptics
+        cluster3_users = [] # e.g., Mixed/Neutral voters
+
+        all_cluster_users = []
+
+        for i in range(num_users_per_cluster):
+            user_id = str(uuid.uuid4())
+            username = f"cluster1_user_{i+1}_{uuid.uuid4().hex[:4]}@example.com"
+            con.execute("INSERT INTO users (id, username) VALUES (?, ?) ON CONFLICT (id) DO NOTHING", [user_id, username])
+            cluster1_users.append(user_id)
+            all_cluster_users.append(user_id)
+
+            user_id = str(uuid.uuid4())
+            username = f"cluster2_user_{i+1}_{uuid.uuid4().hex[:4]}@example.com"
+            con.execute("INSERT INTO users (id, username) VALUES (?, ?) ON CONFLICT (id) DO NOTHING", [user_id, username])
+            cluster2_users.append(user_id)
+            all_cluster_users.append(user_id)
+
+            user_id = str(uuid.uuid4())
+            username = f"cluster3_user_{i+1}_{uuid.uuid4().hex[:4]}@example.com"
+            con.execute("INSERT INTO users (id, username) VALUES (?, ?) ON CONFLICT (id) DO NOTHING", [user_id, username])
+            cluster3_users.append(user_id)
+            all_cluster_users.append(user_id)
+
+
+        # --- Insert comments and associated users ---
+        comment_id_map = {} # Map comment text to comment ID
+        for comment_text in comments_list:
+            comment_id = str(uuid.uuid4())
+            # Generate a random user ID and username for the comment author
+            author_user_id = str(uuid.uuid4())
+            author_username = f"author_{uuid.uuid4().hex[:8]}@example.com"
+
+            # Insert the author user
+            con.execute("""
+                INSERT INTO users (id, username)
+                VALUES (?, ?)
+                ON CONFLICT (id) DO NOTHING
+            """, [author_user_id, author_username])
+
+            # Insert the comment
+            con.execute("""
+                INSERT INTO comments (id, topic_id, user_id, text)
+                VALUES (?, ?, ?, ?)
+            """, [comment_id, topic_id, author_user_id, comment_text])
+            comment_id_map[comment_text] = comment_id # Store the mapping
+
+        # --- Add Cluster Votes ---
+        # Define comment categories based on the example topic context (Civic Tech Initiative)
+        # This is hardcoded based on the context provided in the prompt
+        pro_tech_comments = [
+            "Finally! A system to track rebel scum more efficiently. This will be a glorious day for the Empire!",
+            "Anything that improves the speed of selling junk is good in my book. Maybe I can finally get a decent price for this thermal detonator...",
+            "Fascinating! I am programmed to be compliant. I shall analyze this initiative and report my findings to the Emperor.",
+            "This is a welcome step towards greater efficiency and transparency... cough... as long as it doesn't affect my personal interests.",
+            "As long as it helps me track down my targets, I'm in. The more data, the better.",
+            "The Emperor's vision is one of unparalleled order and prosperity! This initiative will usher in a new era of galactic harmony!",
+            "I'm interested... Will it help me collect debts more efficiently?",
+            "If it improves the entertainment options on Coruscant, I'm all for it.",
+            "Another set of orders. Understood, sir!",
+            "Excellent... with this, I will have even greater control over the galaxy... cackles maniacally"
+        ]
+        anti_tech_comments = [
+            "This is clearly a data-mining operation. They're going to use it to crush the Rebellion. We need to sabotage it!",
+            "The Force guides us to see through their deception. This 'civic tech' will only serve to tighten their grip on the galaxy.",
+            "I just want a reliable power converter. Is that too much to ask? This 'civic tech' sounds like more bureaucracy.",
+            "I'm already dreading the help desk calls. 'My Death Star won't fire!' 'The Force isn't working!'",
+            "This is just a fancy way to track our X-wings. We'll find a way to disable it, just like we did with the Death Star.",
+            "Another reason to drown my sorrows in a Jawa Juice. This whole thing stinks of the Empire's incompetence.",
+            "This initiative is a waste of resources. We should be focusing on military expansion, not 'civic engagement.'"
+        ]
+        # Comments not in pro or anti lists are considered neutral/other for this example
+
+        votes_to_insert = []
+
+        # Cluster 1 (Pro-Tech) votes: Agree with pro, Disagree with anti, Mixed/Neutral on others
+        for user_id in cluster1_users:
+            for comment_text in comments_list:
+                comment_id = comment_id_map.get(comment_text)
+                if not comment_id: continue
+
+                vote_type = None
+                if comment_text in pro_tech_comments:
+                    vote_type = 'agree'
+                elif comment_text in anti_tech_comments:
+                    vote_type = 'disagree'
+                else:
+                    # For neutral/other comments, a chance of neutral or skipping
+                    vote_type = random.choice(['neutral'] * 2 + [None] * 3) # More likely neutral or skip
+
+                if vote_type:
+                     votes_to_insert.append((user_id, comment_id, vote_type))
+
+
+        # Cluster 2 (Anti-Tech) votes: Disagree with pro, Agree with anti, Mixed/Neutral on others
+        for user_id in cluster2_users:
+             for comment_text in comments_list:
+                comment_id = comment_id_map.get(comment_text)
+                if not comment_id: continue
+
+                vote_type = None
+                if comment_text in pro_tech_comments:
+                    vote_type = 'disagree'
+                elif comment_text in anti_tech_comments:
+                    vote_type = 'agree'
+                else:
+                    # For neutral/other comments, a chance of neutral or skipping
+                    vote_type = random.choice(['neutral'] * 2 + [None] * 3) # More likely neutral or skip
+
+                if vote_type:
+                     votes_to_insert.append((user_id, comment_id, vote_type))
+
+        # Cluster 3 (Mixed/Neutral) votes: Mostly neutral, some random agree/disagree, many skipped
+        for user_id in cluster3_users:
+             for comment_text in comments_list:
+                comment_id = comment_id_map.get(comment_text)
+                if not comment_id: continue
+
+                # Mostly neutral, some random agree/disagree, many skipped
+                vote_type = random.choice(['neutral'] * 5 + ['agree', 'disagree'] + [None] * 5) # Weighted towards neutral and skipping
+
+                if vote_type:
+                     votes_to_insert.append((user_id, comment_id, vote_type))
+
+        # Insert all collected votes
+        if votes_to_insert:
+             con.executemany("""
+                 INSERT INTO votes (user_id, comment_id, vote_type)
+                 VALUES (?, ?, ?)
+                 ON CONFLICT (user_id, comment_id) DO NOTHING
+             """, votes_to_insert)
+
+
+        con.commit()
+        # print(f"Successfully added topic '{topic_title}', {len(comments_list)} comments, and {len(all_cluster_users)} cluster users with votes.") # Use print for console output
+        # st.success(f"Successfully added topic '{topic_title}', {len(comments_list)} comments, and {len(all_cluster_users)} cluster users with votes.") # Use st.success if in Streamlit context
+
+    except Exception as e:
+        if con:
+            con.rollback()
+        print(f"Error adding example topic '{topic_title}' and votes: {e}") # Use print for console output
+        # st.error(f"Error adding example topic '{topic_title}' and votes: {e}") # Use st.error if in Streamlit context
+    finally:
+        if con:
+            con.close()
+# Example usage (can be called elsewhere, e.g., in an initialization script)
+def add_dummy_topic():
+    example_topic_id = "15736626"
+    example_topic_title = "New Civic Tech Initiative"
+    example_topic_description = "I want to figure out what the citizens of the Empire really think about the Emperor's new 'Civic Tech' initiative. He's promising streamlined governance, enhanced citizen engagement (apparently), and a 'more user-friendly experience' for everyone. But let's be honest, we all know how the Empire's tech usually works out. So, what are your thoughts? Is this a path to order, or a trap set by the Dark Side? Let's get some honest opinions flowing!"
+    example_comments = [
+        "Finally! A system to track rebel scum more efficiently. This will be a glorious day for the Empire!",
+        "This is clearly a data-mining operation. They're going to use it to crush the Rebellion. We need to sabotage it!",
+        "The Force guides us to see through their deception. This 'civic tech' will only serve to tighten their grip on the galaxy.",
+        "As long as it doesn't mess with my profit margins, I'm indifferent. But I suspect it will.",
+        "I just want a reliable power converter. Is that too much to ask? This 'civic tech' sounds like more bureaucracy.",
+        "Anything that improves the speed of selling junk is good in my book. Maybe I can finally get a decent price for this thermal detonator...",
+        "Fascinating! I am programmed to be compliant. I shall analyze this initiative and report my findings to the Emperor.",
+        "I'm already dreading the help desk calls. 'My Death Star won't fire!' 'The Force isn't working!'",
+        "This is a welcome step towards greater efficiency and transparency... cough... as long as it doesn't affect my personal interests.",
+        "This is just a fancy way to track our X-wings. We'll find a way to disable it, just like we did with the Death Star.",
+        "As long as it helps me track down my targets, I'm in. The more data, the better.",
+        "Another reason to drown my sorrows in a Jawa Juice. This whole thing stinks of the Empire's incompetence.",
+        "The Emperor's vision is one of unparalleled order and prosperity! This initiative will usher in a new era of galactic harmony!",
+        "Will it have cool spaceships in it? Can I play with it?",
+        "Beware the allure of technology. It can be a tool for both good and evil. Trust in the Force, young Padawans.",
+        "This initiative is a waste of resources. We should be focusing on military expansion, not 'civic engagement.'",
+        "I'm interested... Will it help me collect debts more efficiently?",
+        "I'm just trying to survive. This sounds like more trouble than it's worth.",
+        "If it improves the entertainment options on Coruscant, I'm all for it.",
+        "Another set of orders. Understood, sir!",
+        "Excellent... with this, I will have even greater control over the galaxy... cackles maniacally"
+    ]
+    add_example_topic(example_topic_id, example_topic_title, example_topic_description, example_comments)
+
+
 def get_ttl_hash(seconds=360):
     """Return the same value withing `seconds` time period"""
     return round(time.time() / seconds)
@@ -775,7 +980,7 @@ def view_topic_page():
     # Include functional information
     st.markdown(f"**Shareable Quest Scroll ID:** `{topic_id}`")
     # Construct shareable link using current app URL
-    app_url = st.query_params.get('base', ['https://huggingface.co/spaces/npc0/SteamPolis/'])[0] # Get base URL if available
+    app_url = st.query_params.get('base', [DEFAULT_BASE_URL])[0] # Get base URL if available
     shareable_link = f"{app_url}?topic={topic_id}" if app_url else f"?topic={topic_id}"
     st.markdown(f"**Shareable Scroll Link:** `{shareable_link}`")
 
@@ -1081,6 +1286,7 @@ if 'processed_url_params' not in st.session_state:
 
 # Initialize the database on first run
 initialize_database()
+add_dummy_topic()
 
 # Handle initial load from URL query parameters
 # Process only once per session load using the flag